Estimating adoption and impacts of agricultural management practices in developing countries using satellite data. A scoping review

Development and dissemination of sustainable practices are key to enhance agricultural productivity in developing countries and to curtail potential negative externalities. Rigorous adoption/impact evaluations provide valuable lessons to enhance the capacity of agricultural research-for-development (R4D) systems in this context. Conventional evaluation studies rely solely on farm-household surveys for data. Generation of survey data however requires considerable financial and human capital, and the process often misses several important explanatory variables, ignores the longer-term impacts, and suffers from measurement errors. Complementary data sources are explored to make the evaluations more robust and rigorous. Here we review 54 studies that used satellite data to estimate adoption and impact of agricultural practices in developing countries. Some evidence on successful application of satellite data in high-income countries is also provided. The main findings of the paper are threefold: (1) satellite data have been successfully used to detect agricultural practices, such as cropping intensity, tillage, crop residue cover, irrigation, and soil and water conservation; (2) only a few studies have estimated the yield impacts of agricultural practices, although the estimation of crop yields with satellite data is fairly developed; and (3) only a small number of studies have explored impact estimation beyond the biophysical sphere. Estimation of certain environmental impacts of agricultural practices is possible through satellite data, although only a few studies have carried it out. Not many have assessed the economic impacts of interventions. We conclude that satellite data analysis allows information access with little delay and over longer periods, provide a unique set of variables over wide geographies, and reduce measurement error in certain variables. However, more interdisciplinary research is necessary to speed up the uptake of this alternative data source in R4D evaluations.